A consent checkbox is not a consent record

What has to be provable later
When a patient asks who saw their data, the answer has to come from a record that was written at the moment of the exchange and cannot be edited afterwards. A boolean column cannot answer that question; it can only say what the current state is, not how it got there.
Revocation is the hard case
Granting consent is easy to implement and easy to fake. Revocation requires knowing every downstream copy that was authorized by the grant. Recording each grant as an event with an anchor gives the revocation something to point at.
Anchoring, without the theatre
A ledger does not need a token price to be useful. The property that matters is immutability with an audit trail: an append-only history that a third party can verify did not change between two dates. TEFCA's consent model assumes this ledger; the TEFCA IAS page covers the exchange side.